One conversation is all it takes for this AI to deepfake your entire personality
- A new AI model can mimic human personalities after a two-hour chat
- It was 85% accurate in simulating human behavior
- The AI can transform the way behavior and opinions are modeled
Stanford University and Google’s DeepMind have released a new study paper show how they learned an AI model to imitate how humans behave. The Generative agent simulations of 1,000 people Research has created digital replicas of real people that can imitate their personalities with 85% accuracy after just a two-hour interview. Chat with an AI for a few hours, and it will learn to simulate your responses well enough that it looks like it’s thinking just like you.
The just over a thousand participants in the study started by reading the opening lines of The Great Gatsby to the AI. Apparently this was the AI equivalent of stretching before a workout. Then a 2D character asked them questions about their lives, beliefs, jobs, families and more. With about two hours and an average of 6,491 words, the AI had everything it needed to build a digital clone.
The clones also seemed to know what they were doing. When asked to answer questions from personality tests or general surveys, the AI agents provided answers that matched their real-world counterparts about 85% of the time. Close enough to order for you at a restaurant, with perhaps the wrong side dish.
The researchers put these AI clones to the test with economic games such as the Prisoner’s Dilemma and the Dictator Game, where participants make choices about cooperation, trust and sharing resources. Although the AI only matched the real person’s decisions about 60% of the time, that’s still more than can be attributed to chance.
The AI may not fool your partner or best friend, but from that short interview alone it’s pretty impressive. The AI could imitate decision-making patterns, opinions and even personality traits. Think about it: you spend years figuring out who you are, and this AI figures it out in an afternoon. It’s amazing and a little creepy, like all good AI demonstrations.
Human behavior
Stanford and DeepMind see this AI as a way to advance research into human behavior. It’s also a glimpse into how AI can collectively simulate human behavior. Want to know how a community might respond to a new health policy? Set up some generative agents. Curious how customers might respond to a bold product redesign? Ask the AI clones. Essentially, it could be a perpetual focus group.
If an AI can impersonate you so well after one interview, what happens when it has access to years of data? Social media posts, online shopping habits, and even your Spotify playlists can all help the AI learn, or at least accommodate, your preferences, even ones you didn’t know you had.
There is plenty of room for abuse by scammers and other malicious actors, but for now this technology is in the hands of researchers, and the focus is on learning how to make it useful for fields like sociology, psychology and economics. And it is somewhat inevitable that technology will develop in this direction. There is nothing more human than creating something that tries so hard to be like us.